Need your tips to get better prints

[corngolem 0]

Hi all, as I’m new to 3D printing I decided to get a kit to learn how all this work. I’ve assembled it and printed many models with Cura in normal and high quality but the result is not good enough, so I’m requesting your expertise and advice to help me improve it.

I’ve made sure the rods are parallel to each others, I’ve levelled the bed first with a calliper, then positioned the Z top limit switch, then levelled the bed again during a calibration print. I also added nuts to all belt locks except the (machine) left which sounded already tight.

I’ve created an album with pictures of the printed models (with the settings I used), a picture of a bed levelling test print, a video of a model being printed (with sounds) and a video of the belts being rang and the printing head being moved on its axis. You should review those documents before reading further and before answering.

My general observations:

• the worst parts of a print are the bottom, the under (overhang) and the top - I expected the flat bottom and flat top (that is, any flat surface) to be seamless but it was more like a grid

• circles are not round, none of them

• higher speed (100mm/s) seems to be the reason for holes to appear in the print

• the "floors" that are created between infills often do not touch the walls, this makes the print weak and create lots of strings that are prone to creating a destructive blob

• support makes the print worse on other points it’s touching - it might be less true with bigger prints

• below 20% infill, below 10% for sure the slicer is incapable of making floors that bridge (big hole in mesh like roof)

• I levelled the bed with a calliper to 0,05mm precision, yet when I printed I had to unscrew many turns on one side to avoid the nozzle digging in the bed - I don’t understand this - later, printing showed that the perimeter of the bed was levelled but the centre was too high - I don’t see how that’s possible unless the acrylic plate is not flat or one of the rod is not straight

• the high quality setting in Cura quickprint does increase the number of layers but doesn’t seem to give a better result regarding accuracy etc

• the slicer seems to ignore small features even if they are big enough for a 0,4 nozzle

• 0,5mm seems to be missing on all prints for X and Y, no matter the size

• PLA does warp inevitably

• the pathway is often not optimized, when printing a hollow box it would always go back to a starting point instead of going forward and layering up

 

[illuminarti 18]

I think there are a lot of different issues here that are going to take a little while to isolate and work through. But here are some thoughts to consider:

1) You seem to be printing a lot of very small things. That is technically quite challenging, because you need to make sure that lower layers have enough time to cool and solidify before you pile more heat on top. Otherwise, the lower layers will tend to droop and deform. The minimum layer time in Cura helps with this - I generally use a minimum layer time of 7 seconds for PLA which I print at about 220º normally. However, with very small things, slowing down can be counter productive because the head slows down so much that it transfers excess heat into the model anyway, just by lingering over each part. So be sure and set a minimum layer time, but also try printing some slightly bigger parts while you are calibrating things.

2) Circles can end up not round because of backlash in one or other axis, or due to poor layer adhesion resulting in the plastic getting pulled straight. Again, printing some larger parts to begin with will help avoid some of the noise of short, fast moves, head ooze etc, and give a clearer picture of what is really going on. Small prints tend to magnify the impact of small errors, and whats going on can get lost in the random variation of the print. Start by tweaking the printer to print large things well at a 0.2mm layer height, and then once you have that set up well, you can refine things to get equally good results with smaller prints and smaller layers.

3) I don't see any mention of what temperature you are printing at. I'd recommend try a fairly hot temperature - say 230º, and printing at 50mm/s, with 0.2mm layer height, and 0.8mm wall width to begin with. Trying to print too fast can definitely lead to extrusion problems, if you end up extruding too fast for the nozzle. The limit depends on the temperature of the plastic. Printing with those settings should put you in a fairly comfortable middle ground where you can rule out excess speed and insufficiently fluid plastic.

4) Regarding bed levelling. I'm not exactly sure how you are using 'a caliper' to level the bed. Your levelling print looks like the bed isn't level in all the corners, and is too far from the head in most, if not all, of them. The simplest thing to do is to just heat up the nozzle, autohome the z axis, and then manually move the head near to each of the 4 bed screws. Slip a sheet of paper between the nozzle and the bed, and adjust the screw tighter, until the paper can be slid easily back and forth without catching on the head. Then loosen the screw, until the nozzle just begins to catch on the paper. Then loosen the screw another 1/8 turn. Repeat in each corner. Make sure that the underside of each screw head ends up in contact with the bed when you're done. And also make sure that the washers on the springs aren't caught in the keyhole slots (I found it made a huge difference to add bigger washers under the bed).

5) It's possible (but not massively likely) that the bed really is bowed. But I'd follow the process in step 4 first. Then once all four are set, move the head to the middle, and see if you can just slip the paper back under the nozzle with the barest of downward presses on the bed.

6) You seem to be underextruding pretty much everywhere. This is what causes solid surfaces to not fill in properly. This can happen for a number of reasons, including:

a) Using a filament diameter that is less than the diameter stated in the slicer config

b) Printing too fast/thick for the temperature

c) Having your filament feed mechanism be improperly set up. You should have clear deep impressions in the filament as it passes into the bowden tube, but the plastic shouldn't be chewed up or deformed. You might want to try adjusting the tension on the feeder spring. The spring shouldn't be compressed all the way, nor should it be too loose.

d) incorrect steps-per-e setting

7) Slicers are far from perfect. Try different ones, and see what works best for different types of print.

8) Don't bother with the quickprint settings in Cura - go into the full user interface mode, so that you get a proper sense of all the settings that are available, and can begin to tweak things to improve things. The standard modes aren't bad for real-world prints, but are a bit crude for trying to evaluate difficult test prints.

9) Print real world things that you actually like and want - only worry about calibration and test pieces to the extent that they help you print real stuff better. Those test pieces are basically designed to fail at some point; don't worry too much that they do. The only problems that you need to solve are the ones that show up in real prints. :-)

10) Be sure and configure and use retraction. For PLA you will want a distance of 4.5mm, and a speed of around 30mm/s if you have built your own firmware recently, or 40mm/s if you are using the stock firmware from any of the recent Curas. (There is a bug in older firmwares which means that retraction speeds are broken. With fixed firmware you need a speed around 25-35. With the older ones you need a speed of about 40 (which is really 20)). IIRC, the standard profiles in Cura don't use retraction at all, and you seem to be getting a lot of stringing.

11) The reality is that plastic does shrink as it cools, so dimensions are going to be slightly off, and some warping may well occur. With any large, flat, solid pieces that are especially prone to curling, you can minimize the effect of that by wiping down the blue tape with isopropyl alcohol before you start printing. That will stick the base very solidly to the tape, and the tape to the bed. That can't totally avoid warping higher in the print, but it should keep the base firmly stuck.

So, some recommendations....

- Re-level the bed as described above

- Change to the full settings in Cura, and print at the mid-range settings I mentioned... get things sorted out at those settings before pushing the envelope by going faster/thinner.

- Check your filament diameter, and extruder tension to try to fix the under-extrusion issues.

 

- Focus on printing larger objects that you want to print for their own sake. Ultimately the art of 3D printing lies in figuring out how to adjust the settings to best print any given object. To do that, you need to build up experience printing stuff. Calibration prints can drive you nuts. :-)

 

[gr5 2,265]

I agree mostly with illuminatri as usual. The most important thing he mentioned is to tighten that feeder.

Also don't use the canned settings. Look at advanced and expert settings. Know what they all do eventually.

I disagree about the higher temp. For something like the busts, yes go with higher temp. For something like that test piece with all the things sticking up you might want to go lower temp - see this:

http://umforum.ultimaker.com/index.php?/topic/1872-some-calibration-photographs/

If you plan to always print small things because you like small things better then you should also consider going thinner layer height (.1mm instead of .2mm).

The bottom layer is hard to get really good because leveling is so critical. The top layer if flat can be improved by making it thicker (2X or 3X layer height). Always make walls an integer multiple of the nozzle (so .4 or .8) and always make floor/ceiling integral multiple of layer height.

Not sure what you mean about pathway not optimized but I suspect you might want to try kisslicer or the new cura which is coming out soon (in beta now). You might be talking about what we call the "z seam".

"slicer seems to ignore small featuers even if they are big enough for a .4 nozzle" I assume you are talking about what we call "thin walls". If you have a 1mm thick wall in the stl and cura walls are set to .4 then it will leave a .2mm gap. This is a problem fixed in both kisslicer and the "new" cura (in beta). There are current hack methods to fix this - you could lie and say your nozzle is .3mm. But better to wait for the new cura.

 

[corngolem 0]

Some clarifications: I printed small models to get quick results without wasting too much plastic if it failed. I intend to print for other people, so sizes and shapes will be varied.

When I don’t specify the settings it means that the default ones from quickprint normal quality have been used.

By floor I meant not the bottom but an horizontal mesh that is layered between infills.

To illuminatri:

By backslash do you mean that the rods can move inside the frame ? (I noticed this during assembly and added a washer but it didn’t block it) or that a pulley is not held tight to the rod ?

What should be the temperature for 100mm/s and 150 ? is there a formula ?

I use a calliper to measure the distance between over the tape and under the heat block. Right now it’s level to 0,15mm precision.

I already made sure during assembly that the bolts pushed down the plate and that the spring was held by a bottom and top washer.

I don’t know what makes you say my bed isn’t level. If it’s the furthest line on the test print then it always comes out this way, due to early underextrusion I assume. I posted another photo and you should be able to see that the line gets thinner at the center.

I don’t see where it’s under extruding except where there are holes in the 100mm/s prints.

If by underextruding you mean not extruding all the time then I think it’s because the temperature was too low for 100mm/s. If on the other hand you mean that it’s not extruding enough plastic during the whole print then I don’t see the reason for that because I’m using Ultimaking hardware (nozzle, PLA etc) and default software settings (0,4mm nozzle and 2,89mm filament) - unless Ultimaking gave me a different nozzle and a very bad filament (checked, it matches). To me it looks like the pathway is simply not covering the entire surface. It’s doing A instead of B

I didn’t change the steps settings. The filament is held firmly by the feeder, it has marks after it goes through it. However, by design, the feeder is not solidary from the frame - should I secure it ?

What about my belts tension, do they make the right sound in the video ?

I have already tried advanced settings but the current issue don’t let me appreciate them.

To gr5:

In Make magazine’s torture test, when finishing the box, the nozzle would not do sides ABCD, ABCD etc but would go back to a start point.

In Makerbot’s torture test there is a thin wall, I think it should have appeared in the Y=40mm print.

To all:

I printed a full size calibration model (using illuminatri’s settings), took pictures and annotated them to show the defects. Tell me your conclusions.

 

[gr5 2,265]

There's a lot here...

>at about my belts tension, do they make the right sound in the video ?

It did sound fine. But looking at the print - you are right about the walls not touching the infill. Plus your infill is overly sparse (not enough).

There are hints of underextrusion everywhere. The "solid" layers have holes, and the walls don't touch the infill.

This could caused by many things but usually it's that the feeder can't keep up with the nozzle pressure. So, I will repeat what I said earlier: Please tighten the feeder screw that tightens the feeder. Maybe it's fine but give it a full twist and see if that helps anything. Now upping the temp to 240C will reduce head pressure. Alternatively slowing down will help. You can reduce the flow through the head by a factor of 2 by either going with half layer height (.2mm to .1mm) or printing at half speed. Since you are doing experiments, how about printing a 10cm cube solid. Set speed to 100mm/sec. Use ulticontroller to set the speed so that 50% will be 50mm/sec and so on. Start off slow - maybe 20%. After 2 layers double to 40%, 80% 160% and try to see if this makes a difference. You can change it every 2 layers and make notes as you go.

Typically, the printer does fine at 150mm/sec .2mm layers but is actually extruding about 10% less than it is supposed to. But 10% less is not too bad.

You can also play with "flow rate" from the ulticontroller once it starts printing (not before). It's under the calibration menu but doesn't exist until you start printing (from SD card only maybe).

Please post a link to this make magazine torture test. I still dont' understand. I"m guessing the box has 4 points (a,b,c,d) and it always starts at point A and point A is where the z stepper goes up a layer? Is this what you mean? If so that is referred to as the Z seam. Kisslicer has an option to locate the Z seam at random locations each layer. Cura puts it I believe on the near left corner so you can rotate some parts so that the seam is "at the back".

>I printed a full size calibration model

Referred to with underextrusion but there is A SECOND COMMON THING that creates this not-touching-walls issue and that is called "backlash" and just means loose belts. Or worn belts (the teeth). Backlash is where you tell the head to move to the right to position 10, then left back to zero but it only makes it to position 9. A true,pure backlash will not be improved by slower speeds but in reality it might. I had a car where the steering wheel had backlash. I could turn to the right and then when I turned to the left it would move an inch before it started moving the wheels.

But because you seem to have non-solid layers it's probably underextrusion. But you'll have to figure that out.

Finally about levelling. I stared at your leveling photo and it's just too low resolution to tell if the center of your platform is higher so I'll have to trust you. You could try flipping it over maybe. Anyway...

"leveling the bed" has different meanings for different people but usually we mean both levelling *and* setting the home height so that the nozzle is touching the bed. Your leveling print looked level but it looked to low - too much space between nozzle and bed. When you print the very first skirt on a print it should be squished into the blue tape a bit. So I would raise your bed a bit. If you don't get a good first layer it's hard to get everything else right as well. The first layer being bad will create non round circles on *every* layer above. It's hard to believe but very true.

Also the bed will bend over the course of a day and you have to re-level often. With a new printer at least once every 10 hours or maybe more.

With that large test print did you wipe with isopropyl first? Did you get lifting at the corners?

 

[gr5 2,265]

Sorry - I mean 1 cm cube test print. Solid infill. (set to 100%).

 

[illuminarti 18]

Yes, as George said, there's very significant under-extrusion in your full size print, and also really bad backlash causing the infill to not touch the walls.

Under-extrusion just refers to outputting less plastic than needed. The effect can range from gaps between the passes in solid layers, all the way up to holes in vertical walls, or printing cotton-candy fluff - it all depends on the degree to which it happens. It happens for the reasons I explained in my earlier message.

Backlash is when the head doesn't move quite as far as it should before changing direction. It's the major cause of infill not touching walls. The usual reason for this is the short belts not being tight enough. I couldn't tell much from the video about your long belt tension, although I got the impression that they were very tight - tighter than mine, probably. But I'd really need to see the actual belts, rather than just listen to them, given the limitations of web video I think.

Measuring bed height with calipers is over-complicated, and probably difficult to do accurately given the springiness of the bed. Just do the piece of paper thing that I described. The most important thing is that the head is touching the bed, when z=0. That's all that matters. If it's the same height everywhere, that doesn't help you at all if that height doesn't put the nozzle in contact with the bed. Looking at your photo, it does look like the gap between nozzle and bed is slightly less when the head is in the middle. But maybe not enough of a difference to matter. (The difference could be due to the bed not being flat, or due to the 6mm cross rods or axes being slightly bent perhaps?) But what I do notice, as George said, is that the head isn't close enough to the bed at the edges. The first layer of plastic should be slightly squashed and flattened, so as to get good adhesion. In your test print, it looks like the bead of plastic is very round in cross section, and barely attached to the blue tape.

The rods can sometimes slap back and forth if they aren't exactly the right length. A little movement probably won't hurt - it's probably better than too much tension from the end-plates pressing onto them. You might want to try using something like this: http://www.thingiverse.com/thing:54075.

There's no particular formula for temperature vs speed - but in general you'll be able to sustain faster speeds with higher temperature (although may get other effects such as increased stringing).

Are you using the latest Cura? Earlier versions of Cura had a problem when the wall thickness wasn't an exact multiple of the nozzle width that could also cause your 'A v B' under-extrusion-like issues in some circumstances. Make sure your wall thickness setting is an exact multiple of nozzle size, and it won't be an issue either way. But I think that your problem is just that you aren't extruding enough.

You might check the tension in your feeder again... see the photo towards the end of this post for an example of what the teeth marks should look like.

 

[corngolem 0]

Make magazine’s torture test

It doesn’t do it everytime, depends on the settings. It was only an observation anyway, not important.

I don’t have isopropyl so I didn’t use any. If it makes the tape stick to the bed how do you remove the tape ?

I’m using Cura 13.04

The center of the bed is at the proper distance (line is squished into the tape) so I can’t reduce the distance for points on the perimeter. The large scope calibration print stuck well and was accurate so I’m gonna leave it like that.

I tried printing a cylinder at the 4 corners but incredibly the printer started to print from the top ! I tried again to make sure there was no problem with the Z axis, same thing. Try for yourself.

----------------

In my reasoning if there is no underextruding in the walls then there is no underextruding at all. And I don’t see how less plastic can lead to gaps between lines, it already leads to gaps inside the lines.

The feeder’s spring was screwed to measure 13,82mm, I screwed it more to 13,61mm but that did not change the gap between lines. I then unscrewed it to the limit of underextrusion (16,42mm) - that did not change the gap between lines either, but showed underextrusion in the walls. I screwed it back to 13,82 which was when it became harder to screw (as instructions said).

I unscrewed the X&Y motors, found that the spacers were buried in the wood, added washers on that side and screwed them back lower to tighten the belts - (during first assembly I was afraid to put too much strain on the motors).

The result is that the gap between lines almost disappeared.

Then I put longer bolts and more nuts on the long belts lock mechanism - I believe the 4 belts are now ideally tightened.

The result is that the gap is a bit more pronounced than after I tightened the short belts, but less than before. Maybe tightening the long belts was what gave me round circles, I don’t know because I only printed a cube after having tightened the shot belts.

 

I then printed the calibration scope again but with less infill and faster because I didn’t want to wait. The bottoms and top touch the walls more than before but not perfectly everywhere. There was a lot of underextrusion in the walls and I don’t understand what causes it because I printed (different model) walls alone (no infill) at 600% of 25mm/s =150 (walls seem to print slower if no infill, maybe half - couldn’t find the setting) at 230°C and saw no underextrusion.

I suspected retraction because it messes up plastic in the hot end but it doesn’t happen everytime there is retraction. Maybe bad filament quality (from Ultimaking) ?

Could anyone with a fine tuned UM print these gcodes so I can compare with my result and see what I can expect from my printer. (in grey if possible because colours affect the perception of shapes)

 

 

[illuminarti 18]

Solid walls only require that the bead of plastic is the right depth and, within reason, gravity and cohesion will make sure that the small vertical height of each layer is filled in.

Getting solid layers without gaps between the passes requires that the beads are the right width as well so that they touch one another.

However, it looks like you are getting everything sorted out, so I'm sure you'll soon have it all honed to perfection. Enjoy!!

[gr5 2,265]

Ah - so it was the short belts keeping it from touching. I'm glad that makes your tops have fewer gaps but I don't understand how that works - why does less backlash make a solid top?

I really love your pic of the 5 cubes. The one with underextrusion you can see on the side of the cube as well.

On the test photo where you printed up to 600% the corners look worse - that's typical when you don't let a layer cool enough. There is a default setting in cura that slows down the printer if a layer is printed less than 5 seconds to allow the layer to cool. But you defeated this cura feature with your 600% printing. That's why I prefer to always set the mm/sec closer to 100mm/sec so that I can speed it up with UC like you did but max out around 200% (not 600%).

Could you repost that picture of the 5 cubes in this thread here please?:

http://umforum.ultimaker.com/index.php?/topic/1872-some-calibration-photographs/

I'm hoping to create a thread of useful calibration photos and that one is useful. If you want I can do it (if I do it I will move the text to on top of the picture) but you copyrighted it so I am hesitant to steal it.

You are gaining credability with every post and I have to now agree that your plexiglass isn't flat (or your rods aren't straight - less likely). You might want to purchase a piece of glass (maybe 4mm thick?) and clip it to your plexiglass bed using those black paperclips that can clip a 1/2 inch stack of paper (the glass needs to be able to expand if it warms up). That should be a flatter surface.

Be aware that there is a setting in cura in 'expert config' in 'infill' section called 'infill overlap %'. This is designed to fill in that gap between infill and walls. Daid hates it when people increase this to 50% because that is masking the backlash issue (loose belts) but now that you have tightened them nicely I don't mind telling you about it. I increased mine to 20% and am pretty happy with it there. So if you still have a slight gap between walls and infill this is the final tweak to fix it.

 

[corngolem 0]

I’m still discovering but thinking about it I would say that backlash causes gaps in all directions of the XY plane.

The lines not touching the wall is the consequence in one direction of backslash and the lines not touching each other is the consequence (of backslash) in the perpendicular direction.

Is there an inconvenient with a bed made of glass ? why doesn’t Ultimaking use that if it’s guaranteed to be flat ?

Adding it on top of the acrylic bed doesn’t seem the best solution, I would rather replace it with, but gotta find a company who can drill glass...

When I move the print head manually it is smooth on one axis but not on the other. I didn’t know what this was due to, then I thought it was due to the washers I added to secure the rod in the frame, but it might just be that one of them is bent ! (bad rod, bad bed, bad fan - starting to think this printer isn’t worth its price).

I’ve done several tests with speed and temperature, using a 30x5x60mm block with 5% infill, printed at 50mm/s.

At 230°C underextrusion does happen when you reach 500%.

At 240°C and 600% underextrusion never happens, unless retraction occurs. So now I’m pretty sure retraction is the number one cause for underextrusion when the hardware is fine.

I was able to reduce it (by half) by setting retraction distance to 2mm (recommended in Cura) instead of 4,5 (default) - however this created lots of stringing.

I was able to eradicate it by setting retraction extra length on start to 1mm (instead of 0) and leaving distance to 4,5 - but I only tested it on the large scope calibration model.

--------

Thanks to both of you for sharing your knowledge.

 

[illuminarti 18]

I still think you're missing something in your testing which is leading you to conclude that the gaps between passes in the solid layers are a result of backlash. I think something else must be going on. When it's correctly set up, the gcode is exactly enough to exactly fill the available space with nozzle-width strips of plastic. In an enclosed space, that plastic has to go somewhere. While backlash might cause some slight errors in the start and end points of the lines, overall it would have to smear the lines out over say a 50% greater area to create the sort of gaps you saw. And it isn't doing that. So I still think something else is going on. I'm just not sure what.

By definition you should never need the de-retract distance to be longer than the initial retraction, so this definitely indicates that something is going wrong at the extruder end of things. You were seeing this when printing with quite high volume per second, so it may well be that the back pressure that built up in the nozzle was able to shove the filament back extra far during the retraction moves.... as discussed in that article I linked to above.

[gr5 2,265]

illuminarti - please look at his 5 cubes photo and read the text associated with it. It doesn't make sense to me.

 

[gr5 2,265]

The other possibility is that 2 lines were overlapping, 2 lines per layer exactly at the same position in both directions, then they got dissociated when backslash disappeared.

 

That's it! That's exactly what happened. I had to draw it out with pencil and paper. It has to do with the fact that infill is diagonal. If infill was aligned with the axis what you would see is good solid infill that would not touch 3 of the sides.

Instead you have every other line almost on top of each other. Because the plastic has had less than a second to cool the head easily pushes the plastic down to the layer below.

If you don't believe me I can post a diagram. Here's my word explanation (but a picture is better):

To explain you only need backlash on one axis - say the Y axis. I'm going to imagine a vertical square where the infill starts at the top left corner (y=10mm x=0) and diagonally fills until it finishes at the bottom right corner. As each diagonal line hits the left edge, because the Y axis was moving downwards, when it rounds the corner and moves down another line width, it is still moving downwards so the backlash doesn't affect the print and there is a nice .4mm movement to start the next line. But then after moving diagonally up and right and it hits the right edge and the Y axis is commanded to move down .4mm before starting the next run back, it doesn't move at all because it just changed direction and so the next line is on top of the previous (this assumes exactly .4mm backlash - if it's more or less you get slightly different results).

So what you get is a gap every other diagonal infill line. This is great - I learned something new.

 

[corngolem 0]

Yes I want to see your diagram cause I'm not following you entirely. Where does this 0,4mm come from ?

In any case, the question remains: what is creating the gap ?

[gr5 2,265]

The .4mm is the standard distance between diagonal lines if you told cura your head was .4mm. Looking at your pictures of cubes I would say your play aka backlash was around .3 to .4mm in one of the axes. I don't have illustrator or anything really good but I'll give it a shot on this beautiful saturday...

 

[illuminarti 18]

Yes, I think George's explanation makes a lot of sense. New one on me too. Might also explain some of the things I've seen people post lately with weird blobby patterns on their first solid layers... or even ooze in short infill sections, as well. I guess there are really two choices... on the second pass along a line, the extra plastic can either push the lower layer out of the way (if there's room to), just sort of blob our randomly, or mostly get contained in the nozzle, increasing pressure. On starting layers, I would imagine you're mostly going to see either pent-up pressure or random blobbing when the pressure gets high enough to cause oozing out the side.

Well done guys, good detective work!!

On the retraction distance question, bear in mind that the retraction distance is as measured at the extruder gear input. It's very solid plastic at that point, so in theory if the retract and prime distances are the same (and especially if the distances are sufficiently short that the move keeps semi/un-melted plastic within the diameter-constrained hot end, to act as a piston) then you should be exactly back where you started after retracting and re-priming the head. This does ignore the potential for air to get trapped in the head etc. Which is not to say that there aren't valid use cases to have them not be equal to compensate for some other problem - most likely slipping in the extruder gear. I wonder if you weren't especially prone to that, especially during the backlash problems, because you were probably more likely to be getting a pressure build-up in the head, when every other line of plastic had nowhere to go.

 

[gr5 2,265]

Okay, I posted a diagram on the other topic that shows how backlash in either axis will cause gaps in the top infill:

http://umforum.ultimaker.com/index.php?/topic/1872-some-calibration-photographs/?p=14474

In fact if you take it to the extreme the lines can be on top of each other or even be printed in reverse order (every other line on the wrong side of the previous line).

 

[ultifan 0]

Some clarifications: I printed small models to get quick results without wasting too much plastic if it failed. I intend to print for other people, so sizes and shapes will be varied.

When I don’t specify the settings it means that the default ones from quickprint normal quality have been used.

By floor I meant not the bottom but an horizontal mesh that is layered between infills.

To illuminatri:

By backslash do you mean that the rods can move inside the frame ? (I noticed this during assembly and added a washer but it didn’t block it) or that a pulley is not held tight to the rod ?

What should be the temperature for 100mm/s and 150 ? is there a formula ?

I use a calliper to measure the distance between over the tape and under the heat block. Right now it’s level to 0,15mm precision.

I already made sure during assembly that the bolts pushed down the plate and that the spring was held by a bottom and top washer.

I don’t know what makes you say my bed isn’t level. If it’s the furthest line on the test print then it always comes out this way, due to early underextrusion I assume. I posted another photo and you should be able to see that the line gets thinner at the center.

 

I don’t see where it’s under extruding except where there are holes in the 100mm/s prints.

If by underextruding you mean not extruding all the time then I think it’s because the temperature was too low for 100mm/s. If on the other hand you mean that it’s not extruding enough plastic during the whole print then I don’t see the reason for that because I’m using Ultimaking hardware (nozzle, PLA etc) and default software settings (0,4mm nozzle and 2,89mm filament) - unless Ultimaking gave me a different nozzle and a very bad filament (checked, it matches). To me it looks like the pathway is simply not covering the entire surface. It’s doing A instead of B

I didn’t change the steps settings. The filament is held firmly by the feeder, it has marks after it goes through it. However, by design, the feeder is not solidary from the frame - should I secure it ?

What about my belts tension, do they make the right sound in the video ?

I have already tried advanced settings but the current issue don’t let me appreciate them.

To gr5:

In Make magazine’s torture test, when finishing the box, the nozzle would not do sides ABCD, ABCD etc but would go back to a start point.

In Makerbot’s torture test there is a thin wall, I think it should have appeared in the Y=40mm print.

To all:

I printed a full size calibration model (using illuminatri’s settings), took pictures and annotated them to show the defects. Tell me your conclusions.

 

Hi,

In case you still have general under extrusion problems, you could try this,

In Cura, lower filament diameter to approx. 2.85 mm.

Imho, 2.89mm is on the high side of thickness (should be most accurate average thickness of filament)

My experience with Ultimaker filament diameter, varies from min. 2.80 - 2.90 MAX.

So, when you set it to to high, under-extrusion will occur because in reality less filament will extrude based on a too high filament diameter.

Maybe it's also a good idea to reinstall Cura before you change this value...

 

[corngolem 0]

I made 10 measurements on the 3mm Silver PLA from Ultimaking Ltd:

2,9

2,85

3

3,09

2,94

2,96

2,85

2,93

2,83

2,95

Average: 2,93mm (while Cura is set to 2,89 so I should be over extruding)

This might be a factor but I strongly suspect the bowden system. There's a large gap between the filament and the tube, all around, and both are deformed when the head moves around - there's no way this is reliable.

Once I identify the cheapest supplier of PLA 4043D I'll order a roll for the UM and another for the R2 and see what happens.

 

[ultifan 0]

Assuming you have an accurate dial caliper, your filament diam. is quite high... ?!

Indeed, according to your measurements, you should over-extrude.

But, you don't, so something is wrong, just try with 2.85mm and see what happens.

Did you ever changed some steps per E, or density values (should be 1300 ) ?

If so, i should re-install Cura and change to 2.85mm filament diam. and see what happens...

About the gap between your bowden - filament, from what i've seen on many photos this is normal.

Assuming you don't have a faulty / cracked / torn bowden-tube, I don't believe this can cause a problem.

I assume you did install the "blue taped side" of the bowden at the extruder end ?

That's important, because that end of the bowden-tube is widened on the inside, so the filament can enter the tube easily (funnel idea).

Also check if the extruder-tension/ spring properly tightened (but not too tight), about 3-5 turns tight.

 

[corngolem 0]

The calliper is chinese quality, I can measure other parts to verify its accuracy if you tell me which.

I never changed the spets per E.

The gap might be normal but it can't be good because it creates a play.

Actually I put the blue taped end near the feeder because I didn't find any information on where to put it.

If like you say the blue taped end is wider then it should be on the feeder side, not the extruder side.

 

[ultifan 0]

You're absolutely right !

I confused the names extruder and feeder.

The blue-taped end of tube should be at the FEEDER side.

So it can ENTER SMOOTHLY into the bowden like entering a funnel.

 

[ultifan 0]

By the way, i'm not a big fan of chinese made calipers !

Especially NOT, when measuring in 1/100 mm's.

Cheers !

 

[gr5 2,265]

Average: 2,93mm (while Cura is set to 2,89 so I should be over extruding

 

Under. It thinks the filament is bigger than it is so it sends out less filament than it should.

Although the difference between 2.93 and 2.89 shouldn't matter. On the other hand you are a bit picky on quality (which is a good thing) so try increasing flow a bit on the ulticontroller "tune" menu. Or just type in 2.89 for the filament. 2.93 versus 2.89 is a 3% difference (you have to square the ratio as area is square of diameter). But I'd try increaseing flow by 10% if you are seeing gaps.